Abstract

Sportlogiq data can tell us which defensemen play most aggressively (pinch) in the offensive zone. I hypothesize that the defensemen who pinch most frequently have a positive impact on 5v5 offensive shot rates, outweighing the risk of an odd-man rush or zone exit.

To quantify aggressiveness, I include all passes, receptions, shots, loose puck recoveries, and puck protections by defensemen in the offensive zone. I define a pinch as any defenseman’s play that goes closer to the net than the faceoff circles.

My idea is inspired from watching Dan Boyle play behind the net while in the offensive zone with the Tampa Bay Lightning and New York Rangers.

Visualizing Offensive Aggressiveness by Defensemen

To evaluate offensive aggressiveness by defensemen, I define 3 metrics using the aforementioned offensive zone events:

  1. Individual Usage - Percentage of all O-Zone events attributed to one player
  2. Individual Pinch Contributions - Of all of his team’s on-ice O-Zone plays above the circles, what percentage of them are by him?
  3. Individual Pinch Percentage- Percentage of a defenseman’s offensive zone events that are a pinch

I investigate the aggressiveness of Mitchell Vande Sompel, an offensive defenseman. By my new metrics, Vande Sompel is the primary player on 14% of Bridgeport’s offensive zone events when on ice, 8% of Bridgeport’s plays in the pinch area, and 35% of Vande Sompel’s offensive zone events are a pinch.

Below are offensive zone event contours for Bridgeport’s defensemen with 20 games played. All defensemen have a lot of plays at the points, but Vande Sompel’s plot stands out in his pinching all over the offensive zone.

Conversely, Dallas Eakins’ San Diego Gulls do not have any defenders that pinch as frequently as Sebastian Aho (defenseman) or Mitchell Vande Sompel.

Looking at Vande Sompel’s offensive zone events, a good portion of his pinches are shots inside the circles, with some passes, loose puck recoveries, and receptions along the wall. At a glance, he is mostly succesful on his offensive zone passes.

For comparative purposes, we can look at Vande Sompel against the AHL’s most frequent pincher, Reece Willcox (Lehigh Valley Phantoms), and least frequent pincher, Sergei Boikov (Colorado Eagles). Willcox pinches heavily favoring his right-handedness to the right of the faceoff dot, with a lot of action behind the net, but Vande Sompel has a higher frequency of events in between the dots. What you can tell from a glance is that there are a lot of X’s towards the slots, alluding to the difficulty of making plays in the slots. Reece Willcox also has a noticeable amount of green X’s representing failed passes, so I will investigate his passing ability to see if he is hurting his team.

Here are the top 10 players in terms of Pinch Percentage in the AHL:

Player Team iEventsPinch iEvents iPinchPct
Reece Willcox LV 242 555 43.6%
Doyle Somerby CLE 238 633 37.6%
Niko Mikkola SA 223 594 37.5%
Riley Stillman SPR 171 458 37.3%
Mark Friedman LV 241 653 36.9%
Dennis Gilbert RFD 218 601 36.3%
Mark Alt COL 175 484 36.2%
Kurtis MacDermid ONT 120 336 35.7%
Nicolas Meloche COL 168 481 34.9%
Jordan Subban TOR 137 395 34.7%

And 10 least frequent:

Player Team iEventsPinch iEvents iPinchPct
Sergei Boikov COL 38 338 11.2%
Duncan Siemens MIL 40 327 12.2%
Jake Bischoff CHI 70 473 14.8%
Mitch Reinke SA 79 518 15.3%
William Borgen RCH 91 585 15.6%
Brett Lernout LAV 86 530 16.2%
Dylan Coghlan CHI 69 420 16.4%
Chris Casto BRI 61 370 16.5%
Libor Sulak GR 80 485 16.5%
Kyle Burroughs BRI 92 539 17.1%

To see if this is randomness, I look at the cumulative frequency by offensive zone event, and it does appear that the metric does start to stabilize after some time. With more than half a season of data, it may appear more reliable.

Relationship Between Pinch Metrics and Possession Metrics

I look at the aforementioned pinching metrics at the individual and team levels to see if aggressiveness is correlated with expected goals or possession metrics. Corsi and Fenwick are adjusted for score, zone, and venue using Corsica-Hockey’s calculations. At the player level, individual offensive zone usage has the strongest relationship between CF%, FF%, and xGF%, also containing a negative correlation with CA60 and xGA60. However, Pinch % and Pinch Contribution % have no correlation with most shot metrics. I expected the correlations to be small, as pinches make up only 6% of all offensive zone events and and 10% of events below the faceoff circles.

At the team level, I look to see if offensive zone usage of defensemen and percentage of plays below faceoff dots by defensemen have any relationship with possession metrics. The most notable result is a moderately-strong negative relationship between Pinch Contribution Percentage and xGF%. While this is interesting, correlation does not imply causation, so this raises the question if teams that pinch more frequently are using bad strategy, are having their defensemen compensate for a lack of forward talent, or if teams are strategically putting defensemen out of their comfort zone to grow at the AHL level. This is an area for future research.

Clustering Offensive Zone Styles

To compare defensive aggressiveness, I cluster defensemen based on Usage, Pinch Contribution %, Pinch %, Shots Per O-Zone Touch (Shot Happiness), and Passes Per O-Zone Touch (Pass Happiness). Using K-Means on the 174 defensemen with over 300 minutes, I find 4 clusters of defensemen.

Cluster n Usage Pinch Cont % Pinch % Shot Frequency Pass Frequency
Heavy Pinchers 42 13.00% 6.79% 33.0% 13.83% 28.47%
Puck Movers 35 12.19% 4.57% 23.2% 13.18% 29.69%
Low Usage Point Defensemen 42 11.88% 3.58% 18.7% 18.94% 24.62%
Shot-Happy Defensemen 55 13.09% 5.57% 26.3% 17.44% 25.53%

Heavy Pinchers are characterized by their high pinch contributions and percentage, and since they touch the puck so much they have a relatively low frequency. The Puck Movers group have similar shot and pass frequency overall to the Heavy Pinchers, but they generally stick closer to the blue line. Low Usage Point defensemen never really pinch, and their shot-happiness looks very high due to the overall low usage. Shot-Happy defensemen make up the largest quantity of defenders, as they shoot the puck frequently, have pinch and usage rates that are relatively high.

When looking at the possession and expected goals stats of these clusters, Shot-Happy defensemen have the highest overall Corsi-For Percentage, driven by a high Corsi-For Per 60, with the other clusters being very similar. Puck movers don’t have the best CF%, but they have a higher Expected Goals Per 60 than all of ther groups. The Heavy Pinchers group does have the highest XGA60 and the second highest CA60 of the group. This goes against my hypothesis that heavy pinchers outweigh the risks of chances the other way.

Cluster n Pinch % xGF60 xGA60 xGF% CF60 CA60 CF%
Heavy Pinchers 42 33.0% 2.43 2.43 49.91% 40.86 41.88 49.44%
Puck Movers 35 23.2% 2.46 2.32 51.46% 39.08 39.89 49.50%
Low Usage Point Defensemen 42 18.7% 2.31 2.42 48.67% 40.72 42.01 49.21%
Shot-Happy Defensemen 55 26.3% 2.45 2.38 50.66% 43.16 41.20 51.11%

While those are only averages, there are players who perform well and poorly in all 4 clusters. The following table has all 174 players with their clusters, pinch, possession, and expected goals statistics. You can filter the below table by cluster and sort by individual stats.

Pinching while leading and trailing

Defensemen stay back in the offensive zone with a lead, and play more aggressively while trailing. Players and coaches still embrace the shell with a lead, despite research dating back to at least 2014 that suggests the defensive shell increases goals against. I hope to visualize how this looks for defensemen pinching. The metric I use to measure this is Pinch Contribution Percentage, as I want to see if defensemen are making more of a team’s plays in below the circles when trailing.

At the team level (two defensemen on the ice at a time), defensemen generally pinch more frequently in the second period, possibly as a result of the longer change. A team leading in the 3rd period has a defenseman involved in a play down low around 8% of the time, about 10% when tied, and 13% when leading, numbers that seem small but add up quickly in a game. I ignore 1st period 3+ goal leads, as they are very infrequent.

Score Diff 1st Period 2nd Period 3rd Period
3+ 9.35% 8.03%
2 8.14% 9.96% 7.77%
1 9.49% 11.2% 7.95%
0 10.2% 10.9% 9.61%
-1 10.7% 11.3% 12.8%
-2 10.7% 11.6% 13.2%
-3+ 11.5% 12.9%

At the individual player level**, I look at how involved defensemen are based on the score differential and their aggressiveness cluster. When trailing, a Heavy Pincher is involved on 7.5% of touches, but cut it back about 2% when up by more than 1 goal. Puck Movers do pinch more when up than when the game is tied, which is an interesting result. Shot-Happy defensemen essentially lower their aggressiveness at a negative linear pace. Low Usage Point Defensemen do see an uptick in activity when trailing, and pinch below 3% of the time when leading by 1.

** At the player level, only one defensemen is of concern at a time. Hence, the pinch rates appear lower. You cannot add the Pinch Contributions of two players to reach the numbers at the team level, since credit is given to both defenders in the team level metric.

Evaluating the AHL’s most active pinchers

I will compare Christian Wolanin, Mitchell Vande Sompel, Niko Mikkola, Matt Donovan, and Reece Willcox, five of the AHL’s most active pinchers. I also include Joshua Mahura, who is a successful defenseman in the offensive zone, but is on a team that essentially never pinches.

Vande Sompel, Mahura, Donovan, and Wolanin all have similar shot assist percentages, while Mikkola and Willcox lag behind. Vande Sompel has the highest completion percentage, but Wolanin sacrifices some of his completions for passes that will lead to shots.

Player n Shot Assist % Pass Completion %
Christian Wolanin 114 36.0% 63.2%
Mitchell Vande Sompel 101 30.7% 76.2%
Matt Donovan 87 29.9% 57.5%
Joshua Mahura 44 29.5% 54.5%
Reece Willcox 92 19.6% 47.8%
Niko Mikkola 55 16.4% 50.9%

I further breakdown passing by pass type, can be found in the appendix of this document. What separates Vande Sompel and Wolanin apart from the other heavy pinchers are their success rates on slot and east-west passes. Wolanin’s most frequent pass-type is a slot pass, and succeeds 17% more than the average AHL defenseman.

Joshua Mahura, who belongs to the Puck Movers cluster, uses his infrequent pinches to make high-probability north and southbound passes. In the limited sample, his slot passes are usually unsuccessful and infrequently lead to shot assists, so I think this indicates that his selectiveness is his optimal.

Willcox clearly has the ability to get to the dirty areas, but is attempting aggressive plays that do not result in shots. While the sample size is not overwhelmingly definitive, the Phantoms should try to simplify his passing game. His over-aggressive style has his xGF60 in the bottom 10th percentile of AHL defensemen. Mitchell Vande Sompel’s scouting report has always backed up the offensive play, and his passing success seems to explain why his xGF% is closer to 50% than his CF% and FF%. Niko Mikkola is a notch below Vande Sompel and Wolanin in terms of overall offensive skill, and it seems San Antonio uses him more aggressively to recover loose pucks and take assisted shots instead of pass aggressively and make plays.

Another component of pinching with a substantial sample in half-season of games is loose puck recoveries, as a player that leaves the point to recover a puck is creating more possessions for his team. Wolanin, Donovan, and Mahura are all opportunistic when recovering loose pucks, as they each create a shot on over 25% of their LPR attempts. Vande Sompel and Mikkola do not recover loose pucks at such a clip, resulting in only 17% and 19% LPR-to-Shot rates, respectively. Willcox’s success rate is pretty high, but he doesn’t create as many shots as Wolanin, Donovan, and Mahura. Overall, Willcox and Mikkola seem to be empty-calorie players when playing aggressively.

Player n LPR to Shot % LPR Success Rate
Mitchell Vande Sompel 76 17.1% 71.1%
Christian Wolanin 72 25.0% 84.7%
Niko Mikkola 69 18.8% 71.0%
Reece Willcox 63 19.0% 79.4%
Matt Donovan 62 27.4% 75.8%
Joshua Mahura 31 32.3% 83.9%

Conclusion: Application to Hockey Operations

This analysis gives a glimpse into a small percentage of plays at the AHL level. Overall, a pretty low percentage of plays in the offensive zone are made by defensemen, and even fewer are below the faceoff circles to get to high danger areas. There is not a positive relationship between an individual player’s aggressiveness and possession metrics, exemplifying that pinching and offensive zone play is only a portion of a defenseman’s responsibilities at even strength. At the team level, there is a negative relationship between pinching and xGF%, which led me to find that defensemen on the trailing team will increase their aggressiveness as a game goes on.

While I provide a framework for evaluating pinching, I must acknowledge that the very best offensive defensemen will usually be at the NHL level, as their skills are scarce and NHL coaches may be willing to deal with their defensive shortcomings. This heavily influences the sample of AHL players, and therefore similar types of analysis remain necessary to see if pinching is an effective strategy at the NHL level.

This framework can help optimize defensive players’ usage in the offensive zone at even strength. The Belleville Senators can look at Christian Wolanin’s skills and gameplan around his aggressiveness. The Lehigh Valley Phantoms can note that Reece Willcox does have a skill to get to dirty areas, however they may work to develop his decision making in tight areas and try to cut down on his dangerous passes. San Antonio can recognize that Niko Mikkola isn’t contributing much to offense when he is pinching, so maybe pinching less wouldn’t hurt his overall game.

Dallas Eakins’ 2018-19 San Diego Gulls were a successful defensive unit at even strength, even with scarce usage of defensemen deep in the zone. I suggest an NHL staff can use Sportlogiq data to examine the 2019-20 offensive zone activity of Josh Manson, Cam Fowler, and Hampus Lindholm under Dallas Eakins’ coaching schemes, and compare that to how they were used under Randy Carlyle and Bob Murray in 2018-19.

Knowing how frequently an opposing player pinches can help a coach plan in the defensive zone. A forward can attack the point aggressively when they see Andrej Suster with the puck, as he’s probably not going to try to maneuver his way down low. A team can expect Reece Willcox to play aggressively up the boards, and they can prepare to poke check him and force dangerous passes, and maybe have a forward cheat the other way for an odd-man rush or breakaway. Christian Wolanin is not trigger-happy, so forcing him to take an outside shot instead of giving him a chance at a slot pass may put him in a difficult situation. John Gilmour is a shot-happy defenseman, but if his team is trailing by two, his pinching level will look more like that of a Heavy Pincher.

Ryan Stimson has written previously about positionless hockey, and I believe the AHL remains an opportunity for NHL teams to experiment with heavily aggressive defensive play, even with the most skilled defensemen likely having a role in the NHL already. I looked briefly into offensive styles mentioned in his article Tactalytics, and plotted teams that play a point-heavy strategy vs a behind-the-net strategy, and found that some teams with both styles do pinch frequently. A team that plays heavily behind the net like Syracuse can have defensemen pinch for one-timers and east-west passing, while Bridgeport’s heavy point play means that a pinching defenseman may have to create on his own.

Acknowledgements and future considerations

The code that I used to draw the hockey rink was from statswithmatt.com’s article “Plotting a hockey rink in R”. The original article can be found at https://www.statswithmatt.com/post/hockey-rink-in-r/.

It is very well possible that defensemen are used as forwards in certain game situations, and I did not 100% account for this outside of basic EDA. Joey Laleggia was the only labelled defenseman removed from analysis, as eliteprospects.com has him listed as a Left Winger.

Directional passing is a very big portion of evaluating passing down low. However, there is not a concrete way to define the target of an incomplete pass using Sportlogiq data. I believe that this information would provide great insight into interior passing if there was a methodology to estimate failed-pass intention.

Looking into player handedness is an interesting extension of this research that may tell us more about defensemen.

Appendix

  1. Event Plots for Players Analyzed

  2. Player Shot Assist and Completion % Table by Pass Type

Player PassType n Shot Assist % Completion % Frequency League Shot Assist % League Completion % League Frequency
Christian Wolanin eastwest 8 75.0% 100% 7.0% 51.7% 81.8% 3.5%
Joshua Mahura eastwest 2 100% 100% 4.5% 51.7% 81.8% 3.5%
Matt Donovan eastwest 3 33.3% 100% 3.4% 51.7% 81.8% 3.5%
Mitchell Vande Sompel eastwest 9 66.7% 88.9% 8.9% 51.7% 81.8% 3.5%
Reece Willcox eastwest 4 0% 75.0% 4.3% 51.7% 81.8% 3.5%
Christian Wolanin north 18 16.7% 83.3% 15.8% 19.7% 65.7% 14.7%
Joshua Mahura north 12 33.3% 66.7% 27.3% 19.7% 65.7% 14.7%
Matt Donovan north 11 27.3% 63.6% 12.6% 19.7% 65.7% 14.7%
Mitchell Vande Sompel north 19 15.8% 94.7% 18.8% 19.7% 65.7% 14.7%
Niko Mikkola north 9 33.3% 77.8% 16.4% 19.7% 65.7% 14.7%
Reece Willcox north 15 13.3% 40.0% 16.3% 19.7% 65.7% 14.7%
Christian Wolanin northoffboards 30 10.0% 50.0% 26.3% 10.9% 54.9% 37.1%
Joshua Mahura northoffboards 6 0% 50.0% 13.6% 10.9% 54.9% 37.1%
Matt Donovan northoffboards 27 18.5% 70.4% 31.0% 10.9% 54.9% 37.1%
Mitchell Vande Sompel northoffboards 25 20.0% 72.0% 24.8% 10.9% 54.9% 37.1%
Niko Mikkola northoffboards 29 10.3% 44.8% 52.7% 10.9% 54.9% 37.1%
Reece Willcox northoffboards 29 10.3% 55.2% 31.5% 10.9% 54.9% 37.1%
Christian Wolanin slot 35 51.4% 48.6% 30.7% 34.2% 31.4% 22.9%
Joshua Mahura slot 14 7.14% 21.4% 31.8% 34.2% 31.4% 22.9%
Matt Donovan slot 33 39.4% 33.3% 37.9% 34.2% 31.4% 22.9%
Mitchell Vande Sompel slot 12 41.7% 41.7% 11.9% 34.2% 31.4% 22.9%
Niko Mikkola slot 8 12.5% 12.5% 14.5% 34.2% 31.4% 22.9%
Reece Willcox slot 25 20.0% 12.0% 27.2% 34.2% 31.4% 22.9%
Christian Wolanin south 19 52.6% 68.4% 16.7% 41.4% 82.7% 14.5%
Joshua Mahura south 7 71.4% 100% 15.9% 41.4% 82.7% 14.5%
Matt Donovan south 9 33.3% 77.8% 10.3% 41.4% 82.7% 14.5%
Mitchell Vande Sompel south 26 38.5% 80.8% 25.7% 41.4% 82.7% 14.5%
Niko Mikkola south 6 33.3% 66.7% 10.9% 41.4% 82.7% 14.5%
Reece Willcox south 10 60.0% 80.0% 10.9% 41.4% 82.7% 14.5%
Christian Wolanin southoffboards 4 25.0% 100% 3.5% 25.9% 79.7% 6.6%
Joshua Mahura southoffboards 3 33.3% 33.3% 6.8% 25.9% 79.7% 6.6%
Matt Donovan southoffboards 4 25.0% 75.0% 4.6% 25.9% 79.7% 6.6%
Mitchell Vande Sompel southoffboards 10 20.0% 70.0% 9.9% 25.9% 79.7% 6.6%
Niko Mikkola southoffboards 3 0% 100% 5.5% 25.9% 79.7% 6.6%
Reece Willcox southoffboards 9 22.2% 88.9% 9.8% 25.9% 79.7% 6.6%